• First data compilations were focusing on stratigraphy
• Dominant research direction (since 2000s)
• Because we now know where fossils are

Spatial structure of sampling

Close et al. (2020)

Latitudinal gradients

Zhang, Shen, and Erwin (2022)

Niche modeling

Saupe et al. (2019)

Range assessment

Darroch et al. (2022)

Understanding extinctions

Penn et al. (2018)

Bioregionalization

Kocsis, Reddin, and Kiessling (2018)

• Most such research use reconstruction data products →
• DEMs, GCMs, ESMs (e.g. Scotese and Wright (2018), Valdes et al. (2017))
• All are built on tectonic models

… the deeper we go in time.

Buffan et al. (2023)

• How does this tectonic model variable create error in ecological results?
• Documentation of tectonic model use can be cryptic. Redo rotations yourself?
• High-level, easy-to-use, fully flexible and robust solution
• Abstraction allows higher complexity!

Integrate tectonic reconstruction into all analytical pipelines

• Use a tectonic model: reconstruct feature collections

Visualize spatial distribution of records

Compare tectonic models (Devonian)

Use derived data products (PALEOMAP, Campanian)

• Most basic components of the model
• Vary depending on the model (e.g. oceanic plates included?)

Example

plates <- reconstruct("static_polygons", age=65, model="MERDITH2021")
plot(plates$geometry, col="gray", border="black")

• NOT paleocoastlines!
• Modern landmassess partitioned with the static_polygons
• Useful for orientation, i.e. where are things on a map?

Example

coast <- reconstruct("coastlines", age=65, model="MERDITH2021")
plot(coast$geometry, col="gray", border="black")

• input: 2-column matrix of present-day geographic coordinates (longitude/latitude)
• The workhorse of rgplates
• Works because of partitioning polygons (default: static_polygons)

# longitude-latitude coordinates
sydney<- c(151.17, -33.85)
montreal<- c(-73.61, 45.52)

# 2-column matrix 
x <- rbind(sydney, montreal)
colnames(x) <- c("long", "lat")
x

           long    lat
sydney   151.17 -33.85
montreal -73.61  45.52

reconstruct(x, age=106.85, model="MERDITH2021")

         paleolong paleolat
sydney    163.5807 -66.8277
montreal  -20.0174  42.8810

• Using paleocoordinates as input, i.e. those derived from spatial calculations in past space
• Tracing evolution ares that have relevance (e.g. sampling)
• Spatial simulations

randomPoints65 <- reconstruct(randomPoints107,
    from=107, age=65)

Vector spatials (sf)

Present-day 10-degree graticules reconstructed (MERDITH2021 at 65Ma)

• Polygon topology issues (lines work)

Raster spatials (terra)

Reconstructed position of present-day topography (ETOPO1, MERDITH2021 at 65Ma)

• GWS (relies in paleocoord. input)

• Joint development since autumn 2024, rgplates is now part of the GPlates suite
• Current feature set prioritized for ecological analyses, but there is lots more, e.g. velocity calculations →

velo <- velocities(age=65, model="MERDITH2021", 
  output="SpatRaster")

Liz Dowding

• Two major modules: online and offline, pros and cons for both
• Shared interface: primary differece is how the tectonic model is specified:
  • online: character string (GWS)
  • offline: platemodel S4 class (GPlates file links)

• Started as part of the chronosphere project (2018)
• Original purpose: provide easier interface to working with deep-time climate model data
• We need paleocoordinates for the climate models… manually? How?

Versioned data access API

• Multiple attempts were made to build R tools on the GWS, none were advanced enough to be used as a dependency.
  • Reproducibility?! - People who struggle with R will not like Docker…
• No control over model version
• Reconstructing large amounts of data, e.g. the entire Paleobiology Database was not possible

• It never became easy to install an R package with a reliable, automatic instance of pyGPlates dependency
• John Cannon suggested that there is an old, abandoned console interface to the GPlates Application
  • 1. R exports files to the hard drive
  • 2. GPlates does its magic
  • 3. R reads stuff back
• Access files?

• All this was part of it, which became a textbook case of feature creep
• Managable until:
  • R spatial evolution (most code needed to be rewritten)
  • Nuss got a data science job
  • Adam became a faculty member and fell for UNIX

Tectonic modelling, R GPlates API ⟹ rgplates

Working with Xiaodong Qin and John Cannon to equalize the interface.

• Tutorials at rgplates’ website: https://gplates.github.io/rgplates/
• Workshop material from PalAss 2024: adamkocsis/palaeogeography_palass2024
• PhD candidates: Analytical Paleo Workshop/School 2025 - Deadline: February 13!
• Undergraduates: Paleobiology Master at FAU

• Hundreds of users, especially who provided feedback and continue to support!
• Deutsche Forschungsgemeinschaft: TERSANE SPex, Ko 5382/2-1